Despite the widespread use of medications during pregnancy, ethical and methodological barriers to clinical trials make observational studies necessary for evaluating medication safety in this population. Observational studies are prone to biases that often limit their validity due to the lack of randomization; integrating genetic information through discordant sibling designs, polygenic scores, and Mendelian randomization can address several confounding issues. However, application of these three approaches in perinatal pharmacoepidemiology has been limited. Complementing traditional designs with these genetically informed research designs can tackle common biases and strengthen causal inference. This paper focuses on applying genetically informed research designs to child outcomes in perinatal pharmacoepidemiology by reviewing various methods, discussing their strengths and limitations, and examining their application to date, as well as considerations for implementing them in future research. Such considerations include the availability of genetic data, the complexity of integrating genetic data with existing epidemiological data, and selection of appropriate genetic instruments for analyses. Incorporating causal inference in perinatal pharmacoepidemiology can ultimately contribute to enhancing safe medication use during pregnancy.
Carson et al. (Mon,) studied this question.